The long-term goal of this project is to determine whether the clonal architecture of AML samples is relevant for clinical outcomes, and whether that information can be translated into clinical testing to predict prognosis. Our group has recently discovered that the clonal architecture of AML samples can be deduced by whole genome sequencing (which defines ali mutations in every case), followed by deep digital sequencing of ail mutations to define their variant allele frequencies (VAFs). Groups of mutations with similar VAFs represent subclones, Ali AML samples have a founding clone, and most contain 1-3 additional subclones that are derived from the founding clone. The dominant clone at relapse, however, is often a subclone that has acquired addition mutations. These data suggest that specific subclones contain mutations that are relevant for altered grov/th properties and/or altered drug sensitivity, which may contribute to refractory disease or relapse. In this project, we will study the clonal architecture of AML genomes, and determine their relevance for ciinical outcomes, via the following specific Aims: Specific Aim 1: We will use custom capture reagents and deep digital sequencing to determine the rate of clearance of individual AML subclones after induction chemotherapy. We will utilize DNA derived from bone marrow biopsies of 89 AML samples that have already undergone whole genome sequencing. gDNA from bone marrow biopsies obtained 2-4 weeks after initiation of therapy will be subjected to custom capture for all known mutations in each genome, and deep digital sequencing will be performed to assess the rate of clearance of the founding clone and all subclones. Clearance patterns will be correlated with clinical parameters to define impact on outcomes. Specific Aim 2: We will use a stromal-based culture system and xenotransplantation to evaluate the clonal architecture of AML samples grown in vitro and in vivo. Our stromal-based culture system allows for the expansion of most primary AML samples for at least 7 days without significantly altered physical properties or clonal drift. We will treat cultured AML cells with a variety of drugs, and responses will be measured using cell cycle assays and deep digital sequencing to define the drug sensitivity of each sample and each subclone. We will also analyze AML cells that expand in immunodeficient mice to determine whether specific subclones have a growth advantage in mice. These data will be used to identify mutations in subclones that may negatively influence prognosis.